1. Field of the Invention
The present invention relates to an image-classifying apparatus, an image-classifying method, and a program operable to classify image information of a still image or a moving image captured by an imaging apparatus, for example, a digital still camera, a camera-built-in cellular phone, a movie camera, and so on.
2. Description of the Related Art
In recent years, the availability of personal imaging apparatuses, which captures digital images, for example, a digital still camera (hereinafter, a digital camera) and a camera-built-in cellular phone, has spread rapidly. As the storage capacity of an imaging apparatus increases, a user thereof can take many images. However, the user cannot see all the images within a short time in many cases. A technique that allows the user to see the images more efficiently is needed.
In order to solve this problem, Japanese patent application Laid-open No. 2003-58867 discloses a method, comprising: automatically classifying a plurality of images into one of a plurality of groups in accordance with information, which is so-called meta information added to each of the plurality of images. Due to this, the user can classify images and can see easily the plurality of images.
To be more specific, time information indicating when the image was captured is extracted. A histogram is created. A horizontal axis of the histogram indicates a time slot when the image was captured. A vertical axis of the histogram indicates how many images were captured at the time slot. Setting borders of a group to a place where no image was captured, thereby one or more groups are defined. A set of image information belonging to one of the plurality of groups is managed utilizing a corresponding folder, thereby classifying the plurality of images into one of the plurality of groups.
Japanese patent application Laid-open No. H10-254746 discloses another method, comprising: classifying an image into one of a set of groups based on positional information indicating where the image was captured, as follows. An imaging apparatus is connected to a cellular phone, and the shooting apparatus acquires a base station ID, with which the cellular phone communicates. In accordance with a base station ID acquired when an image was captured, the image is classified into one of groups and managed.
However, an expected image group of elements that the user thinks belongs to one event and a generated image group of elements into which are classified according to the prior art do not always correspond to each other. In this specification, an “event” means an action, which a user subjectively perceives and recognizes as a bundle of occurrence, such as a home party, a festival, a memorial ceremony, or a trip.
For example, even when the user wants to classify several days of trip as one event “trip to Hokkaido”, since images were practically captured in several days, when the images are classified into by time, the images are classified into two or more groups, although the images should be classified into one group according to the user's cognizance.
When images are classified using information of a place where the image was captured and a user has attended two events held at home and neighborhood on the same day, the user wants to separately classify the images into two events “home party” and “neighborhood”. However, the images may be classified into the same group, because the images have been taken at places near each other.
These problems have been caused because of the cognitive tendency of users. The users tend to feel actions, which happen near neighborhood within relatively short time and in a small range of movement, belong to one event, and actions, which happen far away from his home within relatively long time and wide range of movement, belong to another event.